Every identity vendor leads with a number. "We match 90-something percent of your records." It sounds like the whole game. It isn't even the right game.
A match rate with no context is a marketing artifact. Matched against what source? Under whose rules? With what tolerance for false joins, the silent error where two different people collapse into one profile and quietly corrupt everything downstream? The headline number answers none of this, and the questions it hides are the ones that actually carry enterprise risk.
The more useful frame is this: identity resolution is a control problem. The organizations that get durable value from it are not the ones who bought the highest advertised accuracy. They're the ones who kept control of the rules, the data, and the failure modes.
Three things you should control
1. The matching logic
Whose definition of "the same person" governs your profiles? If the logic is a vendor black box, you can't tune it to your risk tolerance, explain it to a regulator, or adjust it when a source changes. Control means the rules are yours, deterministic where it matters, configurable where context demands it, and legible to the people accountable for the outcome.
2. Where the data lives
Resolution requires assembling identifiers in one place. That place is the most sensitive surface in your entire customer data estate. If it sits in someone else's cloud, you've outsourced your highest risk asset to optimize a number on a slide. Control means resolution happens inside your boundary.
3. The failure modes
No resolution is perfect, and the dangerous error isn't the missed match, it's the wrong one. Over merging fabricates a customer who doesn't exist; under merging fragments one who does. You should be able to see, tune, and audit both. A vendor optimizing for a public match rate is incentivized to hide exactly this.
Accuracy you can't inspect isn't accuracy. It's a number you've agreed to trust. On identity
Why control produces better accuracy anyway
Here's the part the accuracy contest framing misses: control is not the enemy of accuracy, it's the precondition for it. Your data quality, your identifier coverage, and your business context are the inputs that determine real resolution quality. A platform that lets you bring your rules to your data will, over time, beat a higher advertised black box that can't see your context.
- Your rules, your data: resolution tuned to how your customers actually behave.
- Inspectable joins: every merge explainable, every threshold adjustable.
- Source aware matching: different confidence for a wallet, a loyalty ID, and a web cookie.
How we talk about it. Binoban doesn't publish a fixed identity accuracy figure, because an honest one doesn't exist independent of your data. Resolution quality depends on source quality, identifier coverage, and matching logic, so we give you a rigorous, controllable resolution layer and prove its quality against your data in assessment.
The question to ask instead
Next time a platform opens with a match rate, set it aside and ask: Can I see the rules? Can I keep the data? Can I audit the joins? If the answer to all three is yes, accuracy becomes something you can earn and improve. If the answer is no, the number on the slide is the most control you'll ever have, and you're giving it away on day one.
Take the next step